deep-research

Generate cited deep-research reports via parallel web searches and Gemini synthesis.

Updated Apr 14, 2026
One-click install
npx skills add https://github.com/sr-ai-dev/sr-harness --skill deep-research-sr-ai-dev
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: deep-research
Source: https://github.com/sr-ai-dev/sr-harness/tree/main/skills/deep-research
Command: npx skills add https://github.com/sr-ai-dev/sr-harness --skill deep-research-sr-ai-dev

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

It helps you produce comprehensive, well-cited research reports from a topic or question without manually planning, searching, validating, and synthesizing sources yourself.

Core Features & Use Cases

  • Parallel research orchestration: Coordinates multiple WebSearch agents and optional browser-explorer agents for JS-heavy or dynamic sites to broaden coverage.
  • Cross-validation and traceability: Collects findings per angle and cross-checks evidence before writing the final report with inline citations.
  • Holistic synthesis: Produces an analysis-style report (executive summary, detailed findings, implications, contrarian views, gaps, and confidence assessment) suitable for decision-making.
  • Modes: Interactive mode requests confirmation with a generated plan; autopilot mode runs end-to-end without stopping.

Quick Start

Use deep research by asking: /deep-research What is MCP protocol?

Frequently Asked Questions about deep-research

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate a cited deep research report from a complex question?▼

The deep research process works by coordinating parallel WebSearch agents and a browser-explorer for dynamic sites alongside a Gemini research pass. It cross-validates collected evidence across multiple perspectives before synthesizing everything into a single cited analysis report.

Can I automate web search and evidence validation for multi-perspective synthesis?▼

Yes, you can automate evidence validation using autopilot mode, which runs end-to-end without stopping. The deterministic orchestration flow executes parallel tool searches, cross-checks findings per angle, and produces a holistic synthesis automatically.

Does this deep research tool work with browser explorer for JS-heavy sites?▼

Yes, this deep research tool works with a browser-explorer component specifically for JS-heavy or dynamic sites. It optionally deploys browser-based extraction alongside standard WebSearch to broaden source coverage and capture dynamically loaded content.

What is the best way to structure a research report with confidence levels and counterarguments?▼

The best way to structure a research report with confidence levels is through holistic synthesis. The system generates an analysis-style format containing an executive summary, detailed findings, implications, contrarian views, gaps, and a final confidence assessment.

Do I need to manually plan search angles before cross-validating sources?▼

No, you do not need to manually plan search angles before cross-validating sources. Interactive mode automatically generates a research plan and requests your confirmation before executing parallel searches and evidence validation.

When should I not use an automated deep research orchestration approach?▼

You should not use automated deep research orchestration for simple factual lookups that lack multiple perspectives. The approach is specifically built for complex questions requiring cross-validation, traceability, and synthesis into a comprehensive cited report.